The experimental method for, and analysis of, more than 300 determinations of air influx rates into two unheated ‘closed rooms’ of opposite aspects each fitted (a) with wall grating, or (b) with flue, are described. The average relative efficacy for all observed wind velocities of (a) of 50 sq.in. to (b) of 15 sq.in., so far as facilitating natural ventilation in the rooms used, is found to be 1·0. The analysis reveals definite and complex relations between influx rate and wind velocity. A correlation is made with data from other sources, principally photographs of the flow forms of air streaming past grounded house models, and a theory is advanced to explain the broad features of the observations. Comparison with similar published work is made, and the fundamental necessity of simultaneous measurement of the speed and direction of the main wind stream is stressed. The advantage of the flue and the importance of openings, other than those in the room concerned, are pointed out.The author wishes to record that the investigation was commenced at the instigation of Mr Dean Chandler, to whom he is indebted for en-couragement to publish the results of the further examination of the data obtained. He desires also to acknowledge the enthusiastic interest in the experimental work which his colleagues of the Physical Laboratory displayed throughout the investigation.The measurements of the wind velocity, so essential to the completeness of the observations, frequently necessitated the scaling of roofs in snow, rain and gale, yet at no time were they neglected.The author thanks Dr E. V. Evans, General Manager and Director of the South Metropolitan Gas Co., for granting permission to publish the paper.